IE48605B1 - Dry etching process using plasma - Google Patents

Abstract

Integrated circuit fabrication, e.g., silicon LSI is expedited by plasma etching in any of a novel class of etchants. Appropriate plasma environments are produced by introduction of halide-halogen combinations as exemplified by BCl3-Cl2.

Description

This invention relates to dry etching processes using plasma which depend primarily on chemical reaction rather than, for example, momentum transfer for their etching action.

Fine resolution device and circuits are fabricated by one or a series of steps each involving lithographic resolution followed by selective treatment of regions of device material, lithography is ordinarily carried out in actinic material which is subsequently developed to result in aperture-delineation. Such patterns serve directly or indirectly to mask material undergoing etching or other processing. Selective etching may be carried out by dry processing,by means of a species produced in a plasma, such procedures being preferred to wet processing where particularly fine resolution is desired.

Such fabrication is used in production of large scale integrated (LSI) silicon circuitry. Use is also contemplated for discrete devices and, for example, other semiconductor technology, integrated optical circuitry and magnetic memories.

LSI circuitry as veil as other high resolution planar structures are generally fabricated through a series of levels. One procedure, common to construction of most such circuits, involves first producing a masking layer within a continuous region of actinic material by 4-8 6 OS selective exposure to radiation followed by solution, development to selectively remove either irradiated or non-irradiated material depending on whether the process is a positive or negative resist process. Such masking layers have served as discrete masks, sometimes with the additional step of replication of the pattern in an underlying layer of some more durable layer such as chromium.

This mask technology now in prevalent use in 10 the fabrication of silicon integrated circuits hae undergone considerable development to the present point at which design rules of a few microns are regularly attainable. Discrete masks so used serve for secondary delineation of patterns in expendable photoresist layers which are developed to serve for actual device processing. Photoresist layers are removed at each processing level to permit fabrication at the next level.

It is generally believed that mask technology will be superceded by a maskless technology (direct processing) to produce finer resolution and higher device density. In accordance with such contemplated procedures, primary rather than secondary delineation will be in expendable resist layers tightly adherent to the device undergoing processing. Such resist layers may be true photoresists or nay use short wavelength radiation.

Begardless of procedures whether mask or maakless and regardless of involved technology, a procedure common to all such fabrication involves selective etching of continuous layers of device-functional material. To date, wet etching, for example by use of aqueous acid media, has found satisfactory use. As resolution needs become more stringent inherent limitations become more significant. liquid media reacting with polycryetalline or amorphous layers result in isotropic etching. The resulting undercutting imposes a limit on spacing.

Increasing miniaturisation has resulted in appreciation of the advantages of dry processing. Etc.hi ng by momentum transfer, for example by ion milling, imparts directionality to material removal and eliminates undercutting. High accelerating fields resulting in energetic particle bombardment of surfaces being processed sometimes causes a nev set of problems. lattice damage at some levels of fabrication may even destroy the device.

On the other hand, dry processing may depend upon plasma assisted reactions. Plasma etching, for example, is dependent upon removal primarily due to chemical reaction of the material to be removed with plasmaproduced «tr-hing species. As in momentum transfer processing, the etched product may be easily removed, for instance by system selection to result in a vapour state reaction product. Plasma assisted etching, however, may be isotropic in behaviour. Anisotropy may result from use of large plasma fields and low pressures, but this may in turn produce intolerable lattice damage as well as rapid resist erosion. As in wet etching, imperfect end point detection complicated by unequal wafer-to-wafer etching may result not only in extreme undercutting, hut bIbo in etching of underlying layers. The latter is alleviated by etching systems with pronounced selectivity for material being etched relative to underlying material.

A variety of materials are encountered in LSI production. For present design rules oxide layers are satisfactorily plasma etched with mixtures of methane and oxygen. The same mixtures are applied to many other materials but selectivity is generally poor. Aluminium or aluminium-rich layers are etched by plasma species resulting from introduction of carbon tetrachloride, hut this is a liquid at room temperature and is thhs difficult to monitor. Heating to increase volatility results in unwanted condensation in cool regions.

A recognised problem with CCl^, that of unreliable initial etching, has been solved by use of BClj (see J. Vgc. S£i. Tephnol.. 14, No. 1 p. 266 (1977)).

A continuing problem in aluminium etching is poor discrimination, due to significant etching of SiOg as well as Si. A further problem, attributed to polymer formation, is evidenced by unwanted etching following exposure to the atmosphere. It is believed that a polymer is formed which is difficult to remove and which reacts with atmospheric moisture to produce HC1 which is responsible for continued etching.

In the invention as claimed an aluminium or aluminium-rich surface is etched using a plasma formed from a gaseous mixture including boron trichloride and chlorine wherein the proportion of chlorine in the mixture relative to the total boron trichloride plus chlorine is in the range 0,1 to 20 percent by volume. By means of the invention a satisfactory etch rate and good etch discrimination with respect to common underlying materiale such as silicon and commonly encountered compounds of silicon and also with respect to common resist materials can be obtained. In preferred forms the process can provide anisotropic etching, reducing undercutting and, under suitable conditions, producing substantially vertical etch walls. Furthermore in preferred rorms the loading effect is reduced and this, together with the rapid initial removal of aluminium oxide layers, makes possible increased uniformity of etching.

The invention can be applied to surfaces of aluminium or of aluminium-rich alloys, such as Al-Cu and Al-Si, which show the etching properties of aluminium.

Such alloys are encountered in silicon LSI circuits and devices, Some embodiments of the invention and some experimental examples will now be described by way of illustration.

Etchant Composition The preferred range of composition, expressed in terms of the two essential precursor ingredients, without regard to carrier, diluent, etc., centres about 5 volume percent Clg, This centre composition is found to be quite desirable from the standpoint of etch rate and selectivity. An overall range in the same terms of from 0.5 percent to 20 percent, and preferably from 0.5 to percent defines a range including compositions which suit most contemplated needs. Increasing amounts of chlorine above the preferred maximum value reeult in increasing etch rate, hut are generally accompanied by a tendency toward undercutting. The tendency toward undercutting may be lessened by increasing power and/or decreasing pressure, but this may giye rise to radiation Induced lattice defects. Chlorine content below 0.5 percent by volume results in decreasing etch rate, well below 1000 Angstroms/min., for given conditions. Such lesser chlorine content may be desired for layer thicknesses of about one micron or less.

Likely unintentional ingredients which may be tolerable include oxygen to 5 percent (larger amounts result in significant resist attack) water below about 1 percent, and carbon dioxide up to several percent.

Other intentional ingredients may be present. Certain of these may serve as simple diluent or carrier. Others may serve to control discharge conditions. Examples of the go former include Ng, He and Ar. While plasmas are observed to be extremely stable, use.of very high power could give rise to confinement which may be alleviated through the Inclusion of such diluents.

End product analysis as well as other observations lead to the conclusion that the primary etchant species is atomic chlorine. BClj or plasma derivative of BCl^ is believed to serve as a recombinant thereby shortening inherent etchant species lifetime.

It is found that such shortened inherent etchant lifetime lessens the loading effect. Preferred composition is discussed with a view to system characteristics as applied to fabrication of structures where layers to be etched are of thicknesses in the micron and submicron range. Again, all compositions otherwise suitable Share the prime attribute of good discrimination relative to resist.

Extremes in composition, as well as intermediate compositions result in such slight resist attack as to be generally undetectable. Similarly, discrimination with 4-8605 respect to underlying material, (e.g., elemental silicon-containing as well as chemically combined silicon) is good for tbs entire range of compositions considered.

Critical parameters considered in determining parameter ranges are (l) etch rate, and (2) etch profile. Although some fabrication procedures may tolerate isotropic profiles, design rules of the order of a very few microns and below, generally give rise to a desire for reduced undercutting. A generally preferred parameter range may be set on the basis of an etch rate ratio in a direction parallel to the surface relative to normal to the surface undergoing etching. For many purposes a suitable ratio may be 1:3 or better with a natural preference for ideal anisotropy equivalent to a ratio approaching zero.

For parameter ranges of discussion etch rate is largely dependent upon pressure within the plasma and power with etch rate decreasing as either of these parameters decreases. For many purposes, an etch rate of about 400 Angstroms/min. ia tolerable. This rate is based on the assumption that the layer thicknesses may be of the order of one or a few thousand Angstroms so the total etch time is a few minutes.

On the basis of the above assumptions, extreme pressure and power limits may be set as „ 48605 2 0.05 torr - 0.6 torr and 0.035 watte/cm - 0.7 watts/cm , respectively. These parameters are in turn interrelated with chlorine content.

It has been found that as etch rate decreases 5 (as pressure or power approaches the minimum indicated), the permitted chlorine range increases. As a corollary, as pressure increases, the permitted maximum chlorine content decreases. Chlorine content is generally within a range of from about 0.5 percent to about 8 percent.

A maximum chlorine content for ideal anisotropy is about 6 percent. All chlorine content percentages are based on total BClj - Cl2 mixture. Chlorine content also enters into considerations of etch rate with increasing chlorine resulting in increasing rate. The minimum of 0.1 percent is also the absolute minimum from the rate standpoint.

It is now possible to specify interrelated parameters to result in desired anisotropy as well as etch rate. From the anisotropy standpoint (from an anisotropy ratio, at least as good as 1:3) chlorine content is from 0.1 percent (suitable from the anisotropy standpoint for any pressure - power value within the range indicated) to a maximum chlorine concentration of about 8 percent for minimum values of pressure with accompanying power and minimum power with accompanying pressure, (0.05 torr - 0.28 watts/cm2 to 0.2 torr - 0.035 watts/cm2) to a maximum of about 0.5 percent for maximum pressure - power (0.6 torr - 0.7 watts/cm2).

Parameter constraints for preferred embodiments arise from a desire for an etch rate of at least about 100 Angstroms/min., as well as etch profile. It is found that the latter imposes a limit of about 800 - 1000 Angstroms/min. with further increase in etch rate resulting in a tendency toward near isotropy.

The etch rate limits correspond generally with pressure - power ranges of 0.1 torr -0.1 watts/cm2 to 0.5 torr - 0.5 watts/cm . Chlorine content corresponding to these two limits are from 0.5 volume percent to 7 volume percent corresponding with the minimum power 8 4-8605 pressure pair to 0.5 volume percent - 2 volume percent corresponding with the maximum.

Experimental work to date permits profile control—in fact permits straight walls—for lower chlorine content. Recombination is concluded to be particularly effective at or in the vicinity of walls.

Si£2£i5l-i2_SS_§££b2$ For LSI fabrication, both in silicon technology end elsewhere, a significant contemplated purpose served by aluminium-rich material ia simply that of electrical conductor. Contemplated alloying ingredients are in minor amounts and are generally concerned with characteristics unrelated to conductivity. So for example, it is known that small amounts of silicon, e.g., up to 5 percent may be included for aluminium circuitry directly in contact with silicon-rich material. The purpose here la to preseturate the aluminium (generally about A1 - 2 percent Si) to prevent dissolution and consequent punchthrough of aluminium to underlying Si-rich layers. There may also be up to five percent of oopper present. Admixture of copper generally at a nominal level of about 4 percent by weight is useful in the fabrication of garnet bubble circuitry. Copper serves to reduce electromigration which, if permitted, may cause deterioration of garnet magnetic properties.

Other Comggsitlpnal, Considerations Under usual conditions, the etch rate ratio of etchants discussed for aluminium-rich alloys relative to generally encountered underlying surfaces is likely to be in the range of 10? 1 to 20? 1. Such underlying surfaces include oxide of silicon however produced, as well as silicon, whether mono or polycrystalline. Discrimination for aluminiumrich material relative to Novolao type resists ie also good, at least 10?1. Zt also appears from other experimental experience that etchant attack on other processing resists (as distinguished from mask asking resists) should be adequate. Xt appears that the only underlying surface materials which may present a problem are those which resemble aluminium, such as alkaline earth metals and other elements or related alloys of atomic members substantially below that of silicon. Even under circumstances where discrimination is not large, other attributes of the process may recommend its use.

Processing Conditions Procedures in accordance with the invention may be carried out in any suitable reactor. Plow patterns which result in uniform etching are of course desirable -10 and our experiments were generally carried out in apparatus providing such flow patterns. Apparatus, in this instance of radial flow design, is described by A. 2. 2einberg in Etching for Pattern Definition (Ξ. G. Hughes and M. J. Band, eds.), The Electrochemical Society, Inc. Princeton, NJ, 1976; and E. G. Poulsen, J. Vac. Soi. Technol.. 14. 266, (1977)« Parallel plate systems comprise pairs of plates contained in a suitable vacuum enclosure. Power commonly in the rf range (e.g. 13·56 megahertz) is applied to the driven plate to initiate and sustain a discharge between the plates, the nondriven of which is ordinarily held at ground potential. The term plasma etching as used here may include a variety of procedures commonly designated otherwise. The only requirement for these purposes is primary removal of surface material to be etched through chemical reaction rather than momentum exchange. Nomenclature variations may arise, for example, in accordance with the relative size of the electrodes and the placement of the wafers (either on the driven or nondriven electrode). In the procedure commonly known as reactive ion etching, the driven electrode is substantially smaller than the counter electrode and the material to be etched is placed on the driven electrode.

In the case of the procedure ordinarily referred to as plasma etching the electrodes are more nearly symmetric and the material to be etched is placed on the nondriven electrode.

Parameters subject to control in these reactors are: etch gas composition, pressure, inlet flow rate, power, interelectrode spacing and substrate temperature. Typical ranges for these parameters are: pressure -0.1-1 torr flow rate - 10-500 SCCM (cm^/min standardised at 1 atmosphere and 25 degrees C); power - 100-3000 watts; electrode spacing - 5-50 millimeters; substrate temperature - 25-250 degrees G.

Examples Apparatus: Experiments were conducted on apparatus of radial flow design as described above. Elate diameter was 17* (43 cm) with a plate spacing of 1* (2.5 cm). Both electrodes are hollow to provide for temperature control, generally by water heating or cooling. Specimens to be etched are supported on the lower electrode which is electrically grounded to the reactor.

The reactor is provided with a liquid nitrogen cold trap intermediate the reactor and the vacuum pump to minimise corrosion of the pump. The trap also acts to condense any water vapour and effectively increases the pumping speed relative to water.

Except where noted experiments were conducted with a single 3 (7.5 cm) wafer supporting a masked layer of aluminium-rich material, the layer having a thickness of from one half to one micron over SiOg. While the examples utilised steam oxidised silicon, other experiments used other forms of silicon oxide with no perceptible change in discrimination.

Procedure: With the chamber still closed, water at a temperature of about 80 degrees C was passed through both electrodes for a period of a few minutes to minimise the likelihood of water condensation during loading. The chamber was then opened and the wafer placed on the bottom electrode; the chamber was closed and the pump set in operation. When a pressure of about 30 microns was II attained, the hot water was replaced by flowing cold water at a temperature of about 25 degrees C. Pumping was continued to a base pressure of a few microns, and the precursor etchant gases were introduced. Introduction of gases was varied between limits so as to result in observed pressures of 100-550 microns (0.1 to 0.55 torr), which, for the apparatus and conditions employed, was equivalent to flow rate limits of from 37 to 180 SCCM. The gaseous mixture was excited to produce a plasma by means of a discharge at 13.56 megahertz and 400-600 watts (equivalent to 0.25 to 0.4 watts per square centimetre). Etch rates were determined to he 300-1000 Angstroms per minute. Endpoint was generally detected by an increase in pressure (due to the larger volume of unreacted etchant). In other experiments, endpoint was determined by emission spectroscopy—by observation of disappearance of the AlCl^ band. Conditions were controlled to result in etch time within the range 5-20 minutes.

Example 1 An aluminium 4 percent copper coating of 4000 Angstroms thickness was etched under the following conditions.

Power - 600 watts Pressure - 0.1 torr Precursor gas mixture - 5 volume percent Cl2 95 percent BClj Plow rate - 39 SCCM Etch rate - 300 Angstroms/min.

Profile - Plat vertical etch wall approximately corresponding with position of initial resist edge.

Example 2 A aluminium - 0.5 percent Cu, 0.75 percent Si, alloy layer of a thickness of one micron was etched under the following conditions: Power - 400 watts Pressure - Θ.35 torr Precursor gas mixture - 2 percent Cl2 percent BClj Plow rate - 189 8CCM Etch rate - 1000 Augstroma/min.

Profile - Plat vertical etch wall approximately corresponding with position of initial resist edge. ,2.

Aluminium 2 percent Si, alloy fibre of 7000 Angstroms thickness was etched as follows: Power - 600 watts Pressure - 0.1 torr Precursor gas mixture - 5 volume percent Gig 95 percent BCl^ Plow rate - 39 SGCM Etch rate - 300 Angstroas/min.

Profile - Plat vertical etch wall approximately corresponding with position of initial resist edge.

The following examples were conducted on the alloy of Example 2 (0.3%1 Cu, 0.75% 8i, remainder Al). Conditions were as set forth in Example 2. Gas composition wae varied as indicated.

Example Volume Etch Bate Profile (expressed as lateral etch depth/ yy.rtLp.gl· 4 1 520 anisotropic-undercut 1:2 5 2 800 flat and vertical 0:1 6 3 930 anisotropic-undercut 1:3 7 4 1000 isotropic 1:1 8 5 1380 isotropic 1:1 9 6 1500 isotropic 1:1 10 10 —1500 isotropic 1:1 11 20 -.1500 isotropic 1:1 The following examples were all conducted with introduced chlorine at 6 volume percent at a pressure of 160 microns (0.16 torr) with an aluminium-rich alloy composition as noted in Example 2. The power was varied and the etch rate and profile noted. 60 5 Example Power Etch Sate Profile (expressed as lateral etch depth/ vertical etch depth) 12 400 390 flat and vertical 0s1 13 500 460 anisotropic-undercut 1:6 14 600 580 isotropic 1:1 The following set of examples was based on volume percent chlorine and 400 watt power with varying pressure: Pressure Etch Sate Example (Microns) (A/min.) ProfiXe 15 350 1400 Anisotropic 16 200 950 Anisotropic 17 100 417 Anisotropic Besist attack in each of Examples 15» 16» 1? was acceptable for contemplated purposes.

The following set of examples was conducted at a fixed chlorine content of 5 volume percent at a fixed pressure of 0.1 torr and at varying power.

Power (watts) Etch Bate IAZto-bJ... Profile 18 400 320 Anisotropic 19 500 380 Anisotropic 20 600 520 Anisotropic 25 Besist attack in each of Examples acceptable for the most demanding contemplated use.

The conditions of Example 1 were followed with helium introduced to result in a pressure of approximately 200 microns (approximately equal to 50 volume percent helium based on the total introduced gaseous mixture). Etching was isotropic. The etch rate was increased with little effect on the other observed properties.

Claims (8)

1. An etching process in which a surface to be etched, being of aluminium or an aluminium-rich composition having the etching properties of aluminium 5 is exposed to a plasma formed from a gaseous mixture including boron trichloride end chlorine wherein the proportion of chlorine in the mixture relative to the total boron trichloride plus chlorine is in the range 0.1 to 20 percent by volume whereby the surface is 10 etched predominantly by a chemical reaction.

2. A process as claimed in claim 1, wherein the said proportion is in the range 0.5 to 8 percent by volume.

3. A process as claimed in claim 1 or claim 2 15 wherein the surface to be etched is of a aluminium-rich alloy consisting essentially of aluminium and either silicon or copper or both.

4. A process as claimed in claim 3, wherein the alloy consists essentially of from zero to 5 percent 20 silicon, from zero to 5 percent copper and the remainder aluminium,

5. A process as claimed in any of the preceding claims wherein the etching is confined to selected regions of the surface. 25

6. A process as claimed in claim 5, wherein the selected regions are defined by apertures in an overlying masking layer.

7. A process ae claimed in claim 6, wherein the masking layer ie of an organic resist material, 30 6, A process as claimed in any of the preceding claims wherein the pressure and power within the plasma during etching are within the ranges 0.05 torr to 0.6 torr and 0.035 watt/om 2 to 0.7 watt/cm 2 respectively. 35 9. A process as claimed in claim 8, wherein the said pressure and power are within the ranges 0.1 to 0.35 torr and 0.25 to 0.4 watt/cm 2 ,

8. 10. A process substantially as herein described with respect to any of the Examples